�������� ���� �� ���� ��������� ������. ����� ������������� ��������� : http://ecology.genebee.msu.ru/3_SOTR/CV_Terekhin_publ/2009_Crazy_Fred.doc ���� ���������: Mon Mar 16 11:19:30 2009 ���� ��������������: Mon Oct 1 21:18:14 2012 ���������:

Research Article
My mother is crazy: should I stay or should I go?
Short title: Host manipulation and developmental schedule


Abstract
Like many trophically transmitted parasites, the trematode Microphallus
papillorobustus alters the behaviour of its intermediate host, the
crustacean gammarid Gammarus insensibilis, in a way that favours its
vulnerability to definitive hosts (aquatic birds). Parasitized females
still produce eggs, but because juvenile development occurs inside the
female marsupial brood pouch, young gammarids experience the same risk of
predation as their mother until they exit the marsupium. We explored from
both an empirical and a theoretical point of view the idea that developing
juveniles could adjust their developmental schedule in a state-dependent
manner according to the parasitic status of their mother. We predicted that
juveniles from parasitized females should accelerate their development, or
exit the marsupium at an earlier stage, to avoid predation by birds.
Contrary to our expectation, we observed the opposite result, that is,
juveniles from parasitized females exited the marsupial brood pouch
significantly later than did those from uninfected ones. Although this
phenomenon may illustrate a direct or indirect (i.e. environmentally-
induced) cost of being parasitized, a mathematical model highlighted
another less intuitive possibility: although manipulated females should
have an increased probability of being eaten by birds compared to
uninfected ones, they should also have a reduced risk of predation by other
predators with the net result being in fact a reduced risk of dying from
predation. We discuss these results in relationship with current ideas on
host manipulation by parasites in ecosystems.

Keywords: maternal effect, manipulative parasite, predation, amphipod,
trematode, developmental schedule


Introduction
The role of parasites in the evolution of host life-history traits is
a question that has attracted considerable interest in evolutionary ecology
(M�ller 1997). By definition, parasites are costly to their hosts since
they exploit resources that could otherwise be channelled into host growth,
maintenance or reproduction (Price 1980). Direct costs resulting from this
exploitation are a first cause of between-individual or between-population
variations in life-history traits such as fecundity, growth or survival
(see M�ller 1997 for a review, Thomas et al. 2000, Sorensen and Minchella
2001).
Alternatively, changes in host life-history traits following infection
can also be adaptive responses to parasitism (Minchella 1985, Hurd 2001).
Hosts that are unable to resist infection by other means (e.g.
immunological resistance or inducible defences) are favoured by selection
if they partly compensate for the parasite-induced losses by adjusting
their life-history traits. This prediction is now supported by several
theoretical and empirical examples (Minchella 1985, Hochberg, Michalakis,
and de Mee�s 1992, Forbes 1993, Michalakis and Hochberg 1994, M�ller 1997,
Sorensen and Minchella 2001). For instance, parasitized hosts can
adaptively alter their reproductive effort before dying or being castrated
by either enhancing immediate fecundity (Minchella and Loverde 1981) or
reducing age at maturity (Lafferty 1993, Michalakis and Hochberg 1994,
Sorci, Clobert, and Michalakis 1996, Agnew et al. 1999, Fredensborg and
Poulin 2006). Parasitized hosts have also the potential to adjust life-
history traits such as dispersal (Sorci, Massot, and Clobert 1994, Heeb et
al. 1999, Lion, van Baalen, and Wilson 2006), growth schedule (Sousa 1983,
Minchella 1985) and/or sexual behaviour (Polak and Starmer 1998, Adamo
1999).
Beyond selection for responses which alleviate the direct impact of
parasites on infected hosts, another scenario concerns cases of adaptive
transgenerational phenotypic plasticity, in which parents provide their
offspring with phenotypes to cope with, to resist and/or to avoid
infections. For instance, in the European common lizard Lacerta vivipara,
offspring of mothers highly parasitised by mites had higher values of
several fitness components early in life than offspring of parasite-free
mothers or lightly infested mothers (Sorci and Clobert 1995). Parental
infection has also been found to enhance offspring immunity both in
vertebrates (e.g. Hanson 1998) and in invertebrates (Moret 2006). In
addition to parental influences, offspring themselves can in theory
perceive cues correlated to the external parasitic constraints and/or their
consequences, and adjust their own developmental strategies accordingly
(Poulin and Thomas, 2008). Adaptive responses of the progeny in parasitized
individuals can then be the products of natural selection acting on the
parent as well as on the descendant genomes.
Gammarus insensibilis (Amphipoda, Stock 1966) is one of the most
common invertebrate species in the salt-marsh ecosystems from southern
France (Brun 1971). G. insensibilis from southern France lagoons is
frequently parasitized by the trematode Microphallus papillorobustus
(Microphallidae, Rankin 1940) (Helluy, 1981, Thomas et al. 1995). This
parasite has a complex life cycle including snails from the genus Hydrobia
as first intermediate hosts, G. insensibilis as second intermediate hosts,
and various aquatic birds as definitive hosts (Rebecq 1964). M.
papillorobustus is a manipulative parasite for gammarids: infective larvae
(i.e. cercariae) migrate into the amphipod's brain, encyst in the cerebroid
ganglia and then induce strong behavioural alterations in the host (i.e.
positive phototaxis, negative geotaxis and an aberrant evasive behaviour).
Parasitized gammarids are typically found near the surface water (Ponton et
al. 2005a), a behaviour that renders them more susceptible to predation by
small wading birds (definitive hosts of the parasite, Helluy 1981, 1984,
Thomas et al. 1995). Life-history theory suggests that optimal timing for
juveniles to exit the maternal marsupium should be based on the optimal
balance between maximizing growth and minimizing mortality. Because
juveniles are exposed to the same predation risk as their mother during all
the developmental period, we predicted that those developing inside
parasitized females should exit the brood pouch earlier, thus avoiding
avian predation, by one of two ways, either by accelerating their
development or by exiting at an earlier developmental stage. We conducted
an experiment in controlled conditions in which we disentangled the
influences of parasite and microhabitat on the responses displayed by
juveniles. To assess the problem in a broader ecological context, we also
approached the question using a theoretical model.


Material and methods




Experimental study

We designed an experiment to test the hypothesis that embryos of the
crustacean gammarid Gammarus insensibilis adjust their developmental
schedule according to the parasitic status of their mother.

Biological model
The reproductive biology of G. insensibilis (described in Helluy 1981) is
similar to that of the majority of Gammarus species (Sutcliffe 1992). Males
select females close to their moults and guard them until fertilization of
eggs is possible. After insemination, the male generally guards the female
for a few hours before abandoning her. Fertilized eggs then develop in the
female's brood pouch and the full development of embryos to swimming
juveniles lasts about 11-15 days (data for G. insensibilis at 20�C,
Sutcliffe 1992).




Origin and maintenance of specimens

One large sample of pairs of G. insensibilis in precopula mate guarding was
randomly collected during April 2004 in the Thau lagoon (southern France,
43� 25� N, 3� 35� E) following Thomas et al. (1995). Pairs with infected
males were identified in the field through their aberrant surface
behaviour. Assuming that assortative pairing based on infection status
predominated in the field (see Thomas et al. 1995, Thomas, Renaud, and
C�zilly 1996), we expected that pairs captured at the surface of the lagoon
would be comprised infected males and infected females; on the contrary,
pairs captured at the bottom should consist of uninfected males and
uninfected females. In the laboratory (Station M�diterran�enne de
l'Environnement Littoral, S�te), pairs were kept individually in small
plastic cups (2 cm diameter, 5 cm height) in large tanks (diameter 1.5m, 1
m depth) filled with constantly aerated water from the Thau lagoon (18�C,
38�/��), until mating occurred and females moulted. The top and the bottom
of the cups were closed by a plankton net, so water from the tank could
circulate freely through the cups. After insemination, males of each pair
were sacrificed by exposure to - 80�C for a few seconds, measured in length
(from head to tip of telson) and the head was dissected in order to confirm
their parasitic status. Metacercariae of M. papillorobustus are permanent
ovoid cysts (270 x 350(m, Rebecq 1964) located within the amphipod brain
(Helluy 1981). Females were kept if both partners were infected.

Experimental design
Parasitized gammarids are typically found near the surface in open water
whereas non-infected ones are found in the benthic zone, hidden under
algae. Thus, at least, two environmental parameters differ for infected and
non-infected individuals: light and depth. We assessed separately and
jointly effects of depth, lighting and mother's parasitic status on the
developmental schedule of juveniles. For this, we placed infected and non-
infected fertilized females in four different treatments: (i) light-surface
(control for infected females), (ii) light-bottom, (iii) dark-surface and
(iv) dark-bottom (control for uninfected females) (fig. 1). To manipulate
the level of exposure to light, we used transparent tubes and dark opaque
ones, painted black; half of each kind of tube was placed in the surface
and at the bottom (1 meter depth) of a large tank filled with water from
the Thau lagoon. The experiment started with 20 replicates for each of the
treatment. Thus, after insemination, 20 presumed parasitized and 20
presumed uninfected females were placed in each of the four different
categories previously described. The experiment took place in a room
exposed to the natural photoperiod. The cups were examined and cleaned
daily and provided each time with an excess of fish food (Tetra AniMin).
The experiments were finished when the females moulted again. During the
first three days of the experiment females which died were replaced. At the
end of the experiment, all the females were preserved in 70% EtOH(v/v),
measured in length (from head to tip of telson), and dissected in order to
verify their parasitic status.
For each female, we recorded the intermoult duration (in days) and the
day on which the majority of juveniles (more than 50%) were released. Total
number of emerged viable juveniles were counted daily and preserved in 70%
EtOH(v/v). We randomly selected one juvenile among those that emerged on
the peak day as representative of the brood to which it belonged. In order
to determine developmental stage, these representative juveniles were
measured in length (from head to tip of the third metasomal segment) and
the number of articles of both antennae was counted under a
stereomicroscope (Helluy 1981). Length of juveniles was measured on digital
standardized pictures (Olympus Camedia C-5060 wide zoom, 5.1 megapixels,
magnification x4 on a stereomicroscope Olympus SZ61, x45) using Image J�
software. To estimate the measurement error because of the focusing of the
numeric camera, nine photos of the same gammarid were taken and length of
the juvenile's body measured twice per photo. Statistical analyses revealed
that these measures were not significantly different (Kruskal-Wallis ANOVA,
?�=12.6, df=1, p=0.08). Moreover, we verified that the measurement was
sensitive enough to detect size differences between juveniles from
different females. For this, we used 10 juveniles from the marsupium of one
infected female and 10 juveniles from the marsupium of two uninfected ones.
Juveniles of each photo were measured two times, but not consecutively,
yielding an estimated percentage of error of 18.66% (see Bailey and Byrnes
1990).



Data analysis

All statistical tests were performed following Sokal and Rohlf (1981) and
Siegel and Castellan (1988). Data (females' body size, number of juveniles,
intermoult duration and juveniles' size) were normalized (ln transformed).
When data deviated from normality and/or did not fit a normal distribution
after transformation, we used non-parametric statistics instead of
parametric ones. The contribution of different variables to the development
length was derived by multiple-regression procedure (Draper and Smith
1981). Throughout the paper, values given are mean ( S.E. Results were
considered significant at the 5% level.






Results

Females' biological characteristics
The mean body size of females was not significantly different among the
eight different categories (ANOVA, F7,99=1.10, p=0.37, fig. 2A); nor was
the parasitic load of infected females (ANOVA, F3,58=1.71, p=0.18, fig.
2B). As expected, the number of juveniles was positively and significantly
correlated to female size [ln (number of juveniles)=4.27* ln (female size)
- 7.90, N=99, r2=0.60, p<0.0001]. Similarly, the intermoult duration was
positively and significantly correlated with female size [ln (intermoult
duration)=2.29* ln (female size) + 1.60, N=99, r2=0.09, p=0.003]. When
corrected for maternal size (i.e. residual values), the intermoult duration
was significantly different between females from the eight treatments
(ANOVA, F7,99=4,30, p=0.004, fig. 2C). Moreover, the mean number of
juveniles (corrected for maternal size) of infected females was lower
compared to uninfected ones (Mann-Whitney U-test, Z= 4.23, p<0.0001, fig.
2D). The mean number of juveniles was nevertheless not significantly
different among the four experimental categories of infected females
(Kruskal-Wallis ANOVA, ?2=7.4, df=3, p=0.06, fig. 2D) or among the four
categories of non-infected ones (Kruskal-Wallis ANOVA, ?2=1.88, df=3,
p=0.6, fig. 2D).

Juveniles' biological characteristics
Multiple regressions revealed that three variables were significantly
correlated to length of juvenile development (Table 1) : (i) female's
intermoult duration (corrected for female's size, i.e. residuals), (ii) the
interaction between female's intermoult duration (residuals) and female's
parasitic status and (iii) the interaction between female's intermoult
duration (residuals) and the depth, i.e., surface or bottom (Table 1).
Development time of juveniles from infected females was positively and
significantly correlated with intermoult duration (corrected for size) of
mothers [ln (development time)=0.50* residuals ln (intermoult duration) +
2.21, N=58, r2=0.29, p<0.0001, fig. 3], while there was no such correlation
for juveniles from non-infected females [ln (development time)=-0.03*
residuals ln (intermoult duration) + 2.20, N=41, r2=0.006, p=0.62, fig. 3].
Also, infected females had longer intermoult duration than uninfected ones
of equal size (Mann-Whitney U-test, Z= -5.33, p<0.0001, fig. 2C).
Similarly, development time of juveniles from infected mothers was slightly
but significantly longer than that of uninfected females (Mann-Whitney U-
test, Z=-2.51, p=0.01, fig. 4A), a difference that disappeared once the
analysis was corrected for the intermoult duration of the females (Mann-
Whitney U-test, Z=1.22, p=0.22). Finally, the development time of juveniles
was significantly influenced by depth, being slightly longer for females
maintained at the surface whatever the infection status or light conditions
(clear and dark tubes) (Whitney U-test, Z= 2.14, p=0.03). Intermoult
duration of females (infected and uninfected ones) was not affected by the
depth (Whitney U-test, Z= 0.50, p=0.62, fig. 2C).
Concerning the size of juveniles, none of the variables considered in
the multiple regression were significant (supplementary Table A1, fig. 4B).
Maternal infection status did not affect the developmental stage of
juveniles at the exit of the marsupial brood pouch; all juveniles showed
seven articles in their antennae.


The Model

To find the optimal strategy of host behaviour we built an evolutionary
optimisation model (e.g. Sibly and Callow 1985, Stearns 1992, Terekhin and
Budilova 2001) based a trade-off between reproduction and resistance
against infection. The criterion of optimisation (fitness) was the expected
lifetime reproductive output of the host.


We denote the fraction of energy allocated by the gammarid to its
reproduction as R, its probability to survive up to the moment of
reproduction on the bottom as B, and its probability to survive until
reproduction on the surface as S (all other negative reproductive
consequences of being infected are implicitly included in S). Let us denote
as T the probability for the gammarid to stay uninfected. We make this
assumption that the resistance has a cost and that this probability T
depends on the fraction of energy allocated to resist at the infection.
This fraction is equal to 1-R if all energetic needs other than
reproduction and resistance against infection are constant and therefore
not taken into account. This results in a trade-off between reproduction R
and resistance against trematode infection T which can be described by a
"trade-off curve" (e.g. Sibly and Callow 1986), i.e. by some non-increasing
function T on R. In the simplest case, T=1-R, that is, when all the energy
is allocated to resistance (R=0), the probability of infection is 0 and
when no energy is allocated to resistance (R=1), the probability of
infection is 1 (supplementary Table A2).
Based on the above notations and assumptions, we can write the
following equation for the gammarid's fitness, F,

[pic] (1)

where RTB is the gammarid's expected reproductive success on the
bottom, equal to the reproductive effort (R) multiplied by both the
probability of being uninfected (T), and hence staying on the bottom, and
by survival at the bottom (B). Similarly, the second term gives the
expected reproductive success on the surface, which is equal to the
reproductive effort, R, multiplied by both the probability of being
infected (1-T), and hence going up to the surface, and by the survival at
the surface (S). Substituting the trade-off curve equation T=1-R into (1)
we obtain:

[pic] (2)

or [pic]

The evolutionarily optimal fraction of energy for reproduction, Ropt,
is that value of R which maximizes F. A necessary condition for maximizing
F is that the derivative of F on R should be 0. This gives the following
equation for Ropt

[pic]

from which we obtain:

[pic] (3)

We see that the strategy of energy allocation in our model depends simply
on the ratio of the survival on the surface, S, to that on the bottom, B.
Let us consider this result in more detail.

(i) When the surface conditions are lethal for the gammarid (S = 0),
it is optimal for the gammarid to spend one half of its energy for
reproduction, i.e. Ropt = 0.5, and to spend the other half for resistance
against infection (fig. 5A). In this case F = 0.25B. Thus, when S = 0 it is
optimal to accept a 50% risk of infection and hence ultimately be eaten. In
principle, the gammarid could reduce this risk to 0 by allocating all its
energy to resistance against infection; in this case, it would have no
offspring and its fitness would be equal to 0.

(ii) If S is greater than 0 but less than 0.5B, it is optimal to spend
more energy for reproduction, than if S = 0, and the corresponding fitness
is higher for the same value of B. For example, for S = 0.25B, we have Ropt
= 0.667 and F = 0.333B.

(iii) For all S ( 0.5B, we have Ropt = 1 and F = S. Such a significant
preference of the surface over the bottom is explained by the fact that
remaining on the surface requires no energy, so that all the energy can be
spent for reproduction. This strategy can be advantageous even if the
survival at the surface is less than the survival at the bottom; it is
sufficient that survival at the surface be equal to or greater than half
the value of survival at the bottom.

Certainly, all our conclusions are true for the case of the
particular trade-off curve T = 1-R that we have assumed at the beginning.
However, we may expect that, qualitatively, these conclusions will be
similar for other types of curves because all trade-off curves share the
property that each of two variables is a non-increasing function of the
other.

The supplementary Table A3 presents values of fitness that correspond
to different values of bottom and surface survival. It is evident that
fitness increases as bottom and surface survival increase, but of those two
variables, increase in surface survival has the greater effect on increase
in fitness. For example, for B = 0 and S = 100 we have Fopt = 100 whereas
for S = 0 and B = 100 we have only Fopt = 25. The reason is that even
though the bottom survival is high, additional energy is necessary to stay
at the bottom, and this energy is obtained at the expense of reproduction.
We should note, however, that in our model the surface survival takes into
account not only the risk of predation by birds but also the lower quality
of offspring of infected parents.

Figure 5A shows that when the surface survival is at least a half as
great as the bottom survival, it is optimal for all gammarids to go to the
surface. Because this does not occur in reality, we predict that actual
surface survival is much lower than bottom survival. This prediction seems
contradictory to the observed prolongation of the reproductive period on
the surface. To clarify this situation we present the surface survival, S,
in the above model as a product of two independent survivals, S=SpSi, where
Sp is survival taking into account only predation risks and Si is survival
taking into account only infection costs (we include implicitly in Si not
only the direct risk of death because of infection but all other costs of
being infected, particularly, a higher risk of offspring death because of
underdevelopment). For example, if S=0.25 and the costs of predation and
infection are commensurable then Sp and Si will be 0.5.

In the above model, the effects of two components of surface survival
are indistinguishable. However, from the point of view of shortening or
lengthening of reproductive period, their effects are quite different. If
the reproductive period becomes longer then the predation-accounting
survival decreases and, on the contrary, the infection-accounting survival
increases (because of longer period of reproductive care). The effect of
changing reproductive period by a factor of k corresponds to raising Sp to
the power of k, i.e. by changing Sp to Spk. As regards the effect of
changing reproductive period on infection-accounting survival, we may
assume that its increasing with increasing k can be expressed by raising Si
to the power of 1/k, i.e. by changing Si to Si1/k. The overall effect of
changing reproductive period changes SpSi to Spk Si1/ k.

We may always present Sp and Si as Sa and S1-a to give

[pic]

Calculating the derivative of this expression and equating it to zero

[pic]

we obtain the expression for the optimal value of k

[pic]

We see, in particular, that kopt= 1 for a=0.5, kopt <1 for a>0.5 and
kopt >1 for a<0.5. This is illustrated in Figure 5B for S=0.25 and a=3/4, �
and �.

Thus it is optimal to lengthen the reproductive period when the costs
of infections exceed those of predation and shorten it otherwise. If the
both risks are equal then it is optimal not to change reproductive period.



Discussion

A major challenge of life-history theory is to explain and predict the
phenotypic variation of ages and sizes at transitions between life stages
(Roff 1980, Stearns 1992, Berrigan and Koella 1994). Our study of G.
insensibilis suggested that both maternal environment and parasitism by the
manipulative trematode M. papillorobustus can have an effect on offspring
life history traits. However, contrary to our prediction, young gammarids
did not reduce the time spent in the pouch of their infected mothers,
either by accelerated development or premature release. Conversely, our
results suggest that the timing of exit from the marsupial brood pouch is
delayed for parasitized females and for mothers exposed to surface
environment.
The model we built predicted that development time of juveniles from
manipulated females could be lengthened when the costs of infections exceed
those of predation. If the costs of resistance to the infection in the
bottom and the costs of infection in the surface (direct risks of death
because of infection and all other costs of being infected) are equal, then
by reducing development time when they live on the bottom, juveniles may
compensate for the higher risk of predation on the bottom compared to that
at the surface. By manipulating the behaviour of its host and forcing it to
stay in the surface of the water, M. papillorobustus may indirectly protect
it from predation by predators other than aquatic birds (e.g. fish). This
could lead to non-infected amphipods suffering a higher rate of predation
than infected ones. For young gammarids inside the maternal marsupium, the
optimal balance between maximizing growth and minimizing mortality would
then differ between parasitized and unparasitized females. A lower (net)
predation risk of gammarids at the surface could theoretically explain why
juveniles from parasitized females stay longer inside the maternal
marsupial compared to those of unparasitized females.
Although the idea that manipulated hosts are less likely than
uninfected conspecifics to die from predation by non-host predators may be
provocative, it is supported by other studies. These studies show for
instance, that behavioural alterations induced by manipulative parasites
can decrease the probability of predation by certain predators, usually
those that are unsuitable hosts for the parasite (Levri 1998). Our study
suggests that to understand the selective landscape in which manipulative
changes and its evolutionary consequences occur, it is necessary to
consider the manipulated hosts inside the ecosystem. Directs costs and
indirect consequences of being infected can act in opposite directions so
that the net fitness of infected individuals might be similar to or even
greater than that of uninfected ones (Michalakis et al. 1992, Thomas et al.
2000). Further investigations would be nevertheless necessary in our case
to evaluate the true predation rate by both fish and birds in infected and
non-infected G. insensibilis.
Alternatively, non-adaptive mechanisms could account for the longer
development time of juveniles in parasitized individuals. Our study indeed
confirmed that M. papillorobustus impose significant costs to parasitized
females, influencing several aspects of their reproductive biology. For
instance, infected females had a longer intermoult duration compared to
uninfected ones and suffered a significant fecundity reduction (fewer
juveniles) (see also Thomas et al. 1996). The longer development time of
juveniles in parasitized females was observed whatever the experimental
conditions (surface/bottom, light/dark), and was not associated with a
higher differentiation of the juveniles: body size and developmental stage
at the exiting time were similar for all studied females, whatever their
parasitic status. This result is in accordance with the idea that the
progeny of parasitized gammarids requires a longer time to reach the same
size and development stage than do juveniles from uninfected females.
Because M. papillorobustus directly affect host physiological conditions,
we must also consider the possibility that not only the number but also the
quality of eggs produced by parasitized hosts is affected. In that case, an
extended development would be necessary to compensate for the poor quality
of eggs.
The fact that juveniles from uninfected females also have a longer
development time when females are placed at the surface underlines the
significant influence of the environmental conditions on the development,
but it is in accordance with both the adaptive and non adaptive hypotheses
mentioned above. Indeed, we cannot exclude the possibility that surface
conditions could be perceived by juveniles as a signal of reduced predation
risk (i.e. parasitized females) to which they react by changing the exit
date. Alternatively, surface conditions may also be stressful for
juveniles. Additional experiments would be necessary to determine which
variables among those characterising the surface conditions are most
relevant, and how they actually operate to generate the longer
developmental time observed.
Finally, our study also supported the hypothesis that the behavioural
changes seen in this system are a result of manipulation. It has been
recently suggested that instead of directly modifying the behaviour of
their host, certain parasites may select for collaborative behaviour in
their hosts by imposing additional fitness costs in the absence of
compliance (Zahavi 1979, Soler, M�ller, and Soler 1998, Ponton et al.
2005b). Our experimental design allowed us to keep infected females in a
situation of partial (i.e. light or depth) or total (i.e. light and depth)
"disobedience" given what the host is expected to do to favour the
transmission of the parasite. We found that all infected females suffered a
fitness reduction across all experimental conditions, suggesting that the
behavioural change displayed by the gammarid is more the result of true
parasitic manipulation (see also Helluy and Thomas 2003) rather than a
compromise between the host and the parasite strategies.
In conclusion, this study does not support the initial prediction
that juveniles from parasitized females would accelerate their development
to avoid the predation by birds. The opposite result we found is difficult
to interpret as it may illustrate both a parasitic cost and an adaptive
phenomenon, especially if we suppose that the manipulation exerted by M.
papillorobustus ultimately results in a reduced predation rate of
parasitized individuals compared to unparasitized ones.




Literature cited

Adamo SA. 1999. Evidence for adaptive changes in egg laying in crickets
exposed to bacteria and parasites. Ani Behav 57:117-124.

Agnew P, Bedhomme S, Haussy C, and Michalakis Y. 1999. Age and size at
maturity of the mosquito Culex pipiens infected by the microsporidian
parasite Vavraia culicis. Proc R Soc Lond B 266:947-952.

Bailey RC, and Byrnes J. 1990. A new, old method for assessing measurement
error in both univariate and multivariate morphometric studies. Syst Zool
39(2):124-130.

Berrigan D, and Koella JC. 1994. The evolution of reaction norms - simple-
models for age and size at maturity. J Evol Biol 7:549-566.

Brun B. 1971. Variations intrasp�cifiques et sp�ciation chez deux esp�ces
de gammares d'eau saum�tre du groupe Gammarus locusta (Crustac�s,
Amphipodes) [PhD thesis]. Marseille (France) : Universit� de Provence.

Draper NR, and Smith H. 1981. Applied regression analysis. Wiley New York.

Fredensborg BL, and Poulin R. 2006. Parasitism shaping host life-history
evolution: adaptive responses in a marine gastropod to infection by
trematodes. J Anim Ecol 75:44-53.

Forbes MRL. 1993. Parasitism and host reproductive effort. Oikos 67:444-
450.
Hanson LA. 1998. Breastfeeding provides passive and likely longlasting
active immunity. Ann Allerg, Asthma Im 81:523-537.

Heeb P., Werner I, Mateman AC, K�lliker M, Martin WG, Brinkhof MWG,
Lessells CM, and Richner H. 1999. Ectoparasite infestation and sex-biased
local recruitment of hosts. Nature 400:63-65.

Helluy S. 1981. Parasitisme et comportement. Etude de la m�tacercaire de
Microphallus papillorobustus (Rankin 1940) et de son influence sur les
gammares [PhD thesis]. Montpellier (France) : USTL Montpellier.

Helluy, S. 1984. Relations h�tes-parasites du tr�matode Microphallus
papillorobustus (Rankin 1940). III. Facteurs impliqu�s dans les
modifications du comportement des Gammarus h�tes intermediaires et tests de
pr�dation. Ann Parasitol Hum Comp 59:41-56.

Helluy S., and Thomas F. 2003. Effects of Microphallus papillorobustus
(Platyhelminthes, trematoda) on serotonergic immunoreactivity and neuronal
architecture in the brain of Gammarus insensibilis (Crustacea, Amphipoda).
Proc R Soc Lond B 270:563-568.

Hochberg ME, Michalakis Y, and de Mee�s T. 1992. Parasitism as a constraint
on the rate of life history evolution. J Evol Biol 5:491-504.

Hurd H. 2001. Host fecundity reduction: a strategy for damage limitation?
Trends Parasitol 17: 363-368.
Lafferty KD. 1993. The marine snail, Cerithidea cablifornica, matures at
smaller sizes when parasitism is high. Oikos 68:3-11.

Levri EP. 1998. The influence of non-host predators on parasite-induced
behavioural changes in a freshwater snail. Oikos 81:531-537.

Lion S., van Baalen M, and Wilson WG. 2006. The evolution of parasite
manipulation of host dispersal. Proc R Soc Lond B 273:1063-1071.

Michalakis Y, Olivieri I; Renaud F, and Raymond M. 1992. Pleiotropic action
of parasites: How to be good for the host. TREE 7:59-63.

Michalakis Y., and Hochberg ME. 1994. Parasitic effects on host-life-
history traits: a review of recent studies. Parasite 1(4):291-294.

Minchella DJ, and Loverde PT. 1981. A cost of increased early reproductive
effort in the snail, Biomphalaria glabrata. Am Nat 118:876-881.

Minchella DJ. 1985. Host life-history variation in response to parasitism.
Parasitology 90:205-216.

M�ller AP. 1997. Parasitism and the evolution of host life-history traits.
In: Clayton DH, Moore J, editors. Host-parasite evolution: general
principles and avian models. Oxford University Press, New York.

Moret Y. 2006. 'Trans-generational immune priming': specific enhancement of
the antimicrobial immune response in the mealworm beetle, Tenebrio molitor.
Proc R Soc Lond B 273:1399-1405.

Mousseau TA., and Fox CW. 1998. The adaptive significance of maternal
effects. TREE 13(10):403-407.

Polak M., and Starmer WT. 1998. Parasite-induced risk of mortality elevates
reproductive effort in male Drosophila. Proc R Soc Lond B 265:2197-2201.

Ponton F., Biron DG, Joly C, Duneau D, and Thomas F. 2005a. Ecology of
populations parasitically modified: a case study from a gammarid (Gammarus
insensibilis)- trematode (Microphallus papillorobustus) system. MEPS
299:205-215.

Ponton F., Biron DG, Moore J, M�ller AP, and Thomas F. 2005b. Facultative
virulence: a strategy to manipulate host behaviour? Behav Process 72:1-5.

Poulin R. and Thomas F. 2008. Epigenetic effects of infection on the
phenotype of host offspring: parasites reaching across host generations.
Oikos 117:331-335.

Price P. 1980. Evolutionary biology of parasites. Princeton, NJ: Princeton
University Press.

Rebecq J. 1964. Recherches syst�matiques, biologiques et �cologiques sur
les formes larvaires de quelques Tr�matodes de Camargue [PhD thesis].
Marseille (France) : Facult� de Sciences de Marseille.
Roff D. A. 1980. Optimizing development time in a seasonal environment -
the ups and downs of clinal variation. Oecologia 45:202-208.

Sibly RM., and Callow P. 1986. Physiological Ecology of Animals: An
Evolutionary Approach. Blackwell, Oxford.

Sibly R. M. 1989. What evolution maximizes. Functional Ecology 3:129-135.

Siegel S, and Castellan NJ. 1988. Nonparametric statistics for the
behavioral sciences, 2nd edition. New-York: McGraw-Hill.

Sokal RR., and Rohlf FJ. 1981. Biometry, 2nd edition. New-York: WH Freeman
editor.

Soler JJ, M�ller AP, and Soler M. 1998. Mafia behaviour and the evolution
of facultative virulence. J Theor Biol 191:267-277.

Sorci G, Massot M, and Clobert J. 1994. Maternal parasite load increases
sprint speed and philopatry in female offspring of the common lizard. Am
Nat 144(1):153-164.

Sorci G., and Clobert J. 1995. Effects of maternal parasite load on
offspring life-history traits in the common lizard (Lacerta vivipara). J
Evol Biol 8:711-723.

Sorci G., Clobert J, and Michalakis Y. 1996. Cost of reproduction and cost
of parasitism in the common lizard, Lacerta vivipara. Oikos 76:121-130.

Sorensen RE, and Minchella DJ. 2001. Snail-trematode life history
interactions: past trends and future directions. Parasitology 123:3-18.

Sousa WP. 1983. Host life history and the effect of parasitic castration on
growth: a field study of Certhidea californica Haldeman (Gastropoda:
Prosobranchia) and its trematods parasites. J Exp Mar Biol Ecol 73:273-296.

Stearns SC. 1992. The evolution of life histories. Oxford University Press,
Oxford, UK.

Sutcliffe DW. 1992. Reproduction in Gammarus (Crustacea: Amphipoda): basic
reproductive processes. Freshwater Forum 2: 102-128.

Teriokhin AT, and Budilova EV. 2001. Evolution of Life History: Models
Based on Optimization of Energy Allocation. J General Biol 62(4): 286-295.

Thomas F., Renaud F, Derothe JM, Lambert A, de Mee�s T, and C�zilly F.
1995. Assortative pairing in Gammarus insensibilis (Amphipoda) infected by
a trematode parasite. Oecologia 104:259-264.

Thomas F., Renaud F, and C�zilly F. 1996. Assortative mating by parasitic
prevalence in Gammarus insensibilis (Amphipoda): patterns and processes.
Anim Behav 52:683-690.

Thomas F, Verneau O, Santalla F, C�zilly F, and Renaud F. 1996. The
influence of intensity of infection by a trematode parasite on the
reproductive biology of Gammarus insensibilis (Amphipoda). Int J Parasitol
26:1205-1209.
Thomas F., Villa M, Montoliu I, Santalla F, C�zilly F, and Renaud F. 1998.
Analyses of a debiliting parasite (Microphallus papillorobustus, Trematoda)
and its "hitchhiker" parasite (Maritrema subdolum, Trematoda) on survival
of their intermediate host (Gammarus insensibilis, Amphipoda). J Helminth
Soc Wash 65(1):1-5.

Thomas F., Gu�gan JF, Michalakis Y, and Renaud F. 2000. Parasites and host
life-history traits: implications for community ecology and species co-
existence. Int J Parasitol 30:669-674.

Thomas F, Poulin R, Gu�gan JF, Michalakis Y, and Renaud F. 2000. Are the
Pros as well as Cons to being Parasitized? Parasitol Today 16:533-536.

Zahavi A. 1979. Parasitism and nest predation in parasitic cuckoos. Am Nat
113:157-159.


Figure legends
Figure 1: Experimental procedure (?U: uninfected gammarid females, ?I;
infected gammarid females).

Figure 2: Biological characteristics of G. insensibilis females infected or
not by M. papillorobustus according to the different categories of the
experiment (� S.E) (light gray: uninfected; dark grey: infected)

Figure 3: Relationship between the development time of juveniles (ln) and
the residuals of intermoult duration (ln) according to the parasitic status
of females (black: infected, grey: non-infected).

Figure 4: Biological characteristics of juveniles from infected and non-
infected females (+S.E.) (light gray: uninfected; dark grey: infected)


Figure 5: A) Ropt (evolutionarily optimal fraction of energy for
reproduction) as dependent on B (bottom survival) and S (surface survival);
B) An illustration of dependence of Spk Si1/ k on k for S=0.25 and a=3/4, �
and �.

* p<0,05 ** p<0,0001
Table 1. Results of multiple regression analyses on the development time of
juveniles and predictor variables.

|Parameters |Seq SS|df |F |P |
| | | |ratio | |
|Ln female's size |0.018 |1 |0.059 |0.8078 |
|Residuals ln intermoult duration |6.967 |1 |22.211|<0.0001|
|Residuals ln offspring production |0.048 |1 |0.155 |0.6948 |
|Parasitic status |0.059 |1 |0.19 |0.6640 |
|Lighting |0.744 |1 |2.37 |0.1271 |
|Depth |1.176 |1 |3.749 |0.0561 |
|Parasitic status*lighting |0.455 |1 |1.452 |0.2315 |
|Parasitic status*depth |0.164 |1 |0.524 |0.4709 |
|Lighting*depth |0.013 |1 |0.044 |0.8341 |
|Residuals ln intermoult |4.874 |1 |15.54 |0.0002 |
|duration*parasitic status | | | | |
|Residuals ln intermoult |0.113 |1 |0.361 |0.5494 |
|duration*lighting | | | | |
|Residuals ln intermoult |2.934 |1 |9.354 |0.0030 |
|duration*depth | | | | |

Figure 1
[pic]
Figure 2




[pic]


Figure 3



Figure 4

[pic]


Figure 5
A)









B)


Supplementary Tables and figure
Table A1: Results of multiple regression analyses on the juveniles' size
and different variables.

|Parameters |Seq SS |df |F |P |
| | | |ratio | |
|Ln female's size |0.00565 |1 |2.701 |0.1038 |
|Residuals ln offspring production |0.00203 |1 |0.97 |0.3273 |
|Residuals ln intermoult duration |0.0009 |1 |0.426 |0.5155 |
|Development time of juveniles |0.000163|1 |0.078 |0.7804 |
|Parasitic status |0.00205 |1 |0.981 |0.3245 |
|Lighting |0.00000 |1 |0.0002|0.9885 |
|Depth |0.00046 |1 |0.2201|0.6402 |
|Parasitic status*lighting |0.00049 |1 |0.2359|0.6284 |
|Parasitic status*depth |0.00254 |1 |1.2171|0.2729 |
|Lighting*depth |0.000857|1 |0.4097|0.5238 |


Table A2: The values of Ropt as dependent on B and S (when only R is
optimized). (Table probabilities and fractions are given in percents).




Table A3: The values of Fopt as dependent on B and S (when only R is
optimized). (Table probabilities and fractions are given in percents).